U.S. patent application number 11/534345 was filed with the patent office on 2008-03-27 for method, system and computer program product for providing user-customizable standardized anatomical viewing protocols for volumetric data.
Invention is credited to Ingmar Bitter, Doron Dekel.
Application Number | 20080074422 11/534345 |
Document ID | / |
Family ID | 38834474 |
Filed Date | 2008-03-27 |
United States Patent
Application |
20080074422 |
Kind Code |
A1 |
Dekel; Doron ; et
al. |
March 27, 2008 |
METHOD, SYSTEM AND COMPUTER PROGRAM PRODUCT FOR PROVIDING
USER-CUSTOMIZABLE STANDARDIZED ANATOMICAL VIEWING PROTOCOLS FOR
VOLUMETRIC DATA
Abstract
A method, system and computer program product for automatically
generating a set of views for a new set of voxel data according to
a viewing protocol are provided. This involves a) storing a
reference set of views for a reference set of voxel data according
to the viewing protocol, wherein each view in the reference set of
views is defined by a set of geometrical parameters defined with
respect to a reference coordinate system of the reference set of
voxels; b) defining a registration mapping from the reference set
of voxel data to the new set of voxel data; and, c) for each view
in the reference set of views, using the registration mapping to
map each view in the reference set of views to the new set of voxel
data to provide the set of views for the new set of voxel data
according to the viewing protocol.
Inventors: |
Dekel; Doron; (Toronto,
CA) ; Bitter; Ingmar; (Maple, CA) |
Correspondence
Address: |
BERESKIN AND PARR
40 KING STREET WEST, BOX 401
TORONTO
ON
M5H 3Y2
US
|
Family ID: |
38834474 |
Appl. No.: |
11/534345 |
Filed: |
September 22, 2006 |
Current U.S.
Class: |
345/427 ;
345/424 |
Current CPC
Class: |
G06T 2207/30004
20130101; G06T 15/08 20130101; G06T 7/38 20170101 |
Class at
Publication: |
345/427 ;
345/424 |
International
Class: |
G06T 17/00 20060101
G06T017/00 |
Claims
1. A method of automatically generating a set of views for a new
set of voxel data according to a viewing protocol, the method
comprising: a) storing a reference set of views for a reference set
of voxel data according to the viewing protocol, wherein each view
in the reference set of views is defined by a set of geometrical
parameters defined with respect to a reference coordinate system of
the reference set of voxels; b) defining a registration mapping
from the reference set of voxel data to the new set of voxel data;
and, c) using the registration mapping to map each view in the
reference set of views to the new set of voxel data to provide the
set of views for the new set of voxel data according to the viewing
protocol.
2. The method as defined in claim 1 wherein the reference set of
views comprises at least two views.
3. The method as defined in claim 2 wherein step a) comprises
reviewing different views of the reference set of voxel data to
select and store the reference set of views.
4. The method as defined in claim 1 further comprising generating a
set of views for a 2nd new set of voxel data according to the
viewing protocol after generating the set of views in step c), by
d) defining a registration mapping from the reference set of voxel
data to the 2nd new set of voxel data; e) using the registration
mapping to map each view in the reference set of views to the 2nd
new set of voxel data to provide the set of views for the 2nd new
set of voxel data according to the viewing protocol.
5. The method as defined in claim 1 wherein the viewing protocol
further comprises a set of geometrical constraints including at
least one of i) a shape constraint for preserving a view shape
during step c), ii) an angular constraint for preserving a viewing
angle during step c), iii) a spacing constraint for preserving
internal relative spacing during step c), iv) an angular
view-to-view constraint for preserving an angular relationship
between different views in the set of views during step c), and v)
a spacing view-to-view constraint for preserving a spacing
relationship between different views in the set of views during
step c); wherein step c) further comprises transforming the set of
views to comply with the geometrical constraints to generate a
corrected set of views according to the viewing protocol.
6. The method as defined in claim 1 wherein, in step c), the
registration mapping is a volume-to-volume registration
mapping.
7. The method as defined in claim 1 wherein the reference set of
voxel data is obtained from at least one of (i) an anatomical atlas
for storing the reference set of voxel data segmented into labeled
regions; (ii) a person being scanned; (iii) a composite obtained
from a person being scanned multiple times; and, (iv) a composite
obtained from a plurality of people being scanned.
8. The method as defined in claim 1 wherein each view in the
reference set of views is defined by a set of non-geometrical
parameters defined with respect to the reference set of voxels,
wherein the set of non-geometrical parameters comprises at least
one attribute selected from the group consisting of contrast,
transparency and color.
9. The method as defined in claim 8 wherein, in step c), using the
registration mapping to map each view in the reference set of views
to the new set of voxel data comprises, for each view in the
reference set of views, defining a corresponding view in the new
set of voxel data by, i) mapping a geometrical parameter in the set
of geometrical parameters to a corresponding geometrical parameter
of a new set of geometrical parameters defined with respect to a
new reference coordinate system of the new set of voxel data, and
ii) modifying each non-geometrical parameter defined with respect
to the set of voxel data based on the new set of voxel data.
10. The method as defined in claim 1 further comprising, before
step b), selecting the viewing protocol from a plurality of viewing
protocols, wherein step a) comprises storing a plurality of
reference sets of views for a plurality of reference sets of voxel
data, wherein each reference set of views in the plurality of
reference sets of views is of a corresponding reference set of
voxel data in the plurality of reference sets of voxel data, and
defines a corresponding viewing protocol in the plurality of
viewing protocols.
11. The method as defined in claim 10 wherein the plurality of
reference sets of voxel data includes at least one of a male
reference set of voxel data, a female reference set of voxel data,
an adult reference set of voxel data, a child reference set of
voxel data, a first condition reference set of voxel data for a
first physiological condition, and a second condition reference set
of voxel data for a second physiological condition.
12. The method as defined in claim 1 further comprising
automatically selecting the viewing protocol from the plurality of
viewing protocols based on at least one of an identify of a user, a
profession of the user, an organization affiliation of the user,
and a content of the new set of voxel data.
13. A system for automatically generating a set of views for a new
set of voxel data according to a viewing protocol, the system
comprising: a memory for storing the new set of voxel data and the
viewing protocol; means for performing the steps of a) storing a
reference set of views for a reference set of voxel data according
to the viewing protocol, wherein each view in the reference set of
views is defined by a set of geometrical parameters defined with
respect to a reference coordinate system of the reference set of
voxels; b) defining a registration mapping from the reference set
of voxel data to the new set of voxel data; and, c) using the
registration mapping to map each view in the reference set of views
to the new set of voxel data to provide the set of views for the
new set of voxel data according to the viewing protocol.
14. The system as defined in claim 13 wherein the reference set of
views comprises at least two views.
15. The system as defined in claim 14 further comprising a display,
wherein step a) comprises reviewing different views of the
reference set of voxel data on the display to select and store the
reference set of views.
16. The system as defined in claim 13 wherein the viewing protocol
further comprises a set of geometrical constraints including at
least one of i) a shape constraint for preserving a view shape
during step c), ii) an angular constraint for preserving a viewing
angle during step c), iii) a spacing constraint for preserving
internal relative spacing during step c), iv) an angular
view-to-view constraint for preserving an angular relationship
between different views in the set of views during step c), and v)
a spacing view-to-view constraint for preserving a spacing
relationship between different views in the set of views during
step c); wherein step c) further comprises transforming the set of
views to comply with the geometrical constraints to generate a
corrected set of views according to the viewing protocol.
17. The system as defined in claim 13 wherein, in step c), the
registration mapping is a volume-to-volume registration
mapping.
18. The system as defined in claim 13 wherein the reference set of
voxel data is obtained from at least one of (i) an anatomical atlas
for storing the reference set of voxel data segmented into labeled
regions; (ii) a person being scanned; (iii) a composite obtained
from a person being scanned multiple times; and, (iv) a composite
obtained from a plurality of people being scanned.
19. The system as defined in claim 13 wherein each view in the
reference set of views is defined by a set of non-geometrical
parameters defined with respect to the reference set of voxels,
wherein the set of non-geometrical parameters comprises at least
one attribute selected from the group consisting of contrast,
transparency and color.
20. The system as defined in claim 19 wherein, in step c), using
the registration mapping to map each view in the reference set of
views to the new set of voxel data comprises, for each view in the
reference set of views, defining a corresponding view in the new
set of voxel data by, i) mapping a geometrical parameter in the set
of geometrical parameters to a corresponding geometrical parameter
of a new set of geometrical parameters defined with respect to a
new reference coordinate system of the new set of voxel data, and
ii) modifying each non-geometrical parameter defined with respect
to the set of voxel data based on the new set of voxel data.
21. The system as defined in claim 13 further comprising a user
input device for selecting the viewing protocol from a plurality of
viewing protocols, wherein step a) comprises storing a plurality of
reference sets of views for a plurality of reference sets of voxel
data, wherein each reference set of views in the plurality of
reference sets of views is of a corresponding reference set of
voxel data in the plurality of reference sets of voxel data, and
defines a corresponding viewing protocol in the plurality of
viewing protocols.
22. The system as defined in claim 21 wherein the plurality of
reference sets of voxel data includes at least one of a male
reference set of voxel data, a female reference set of voxel data,
an adult reference set of voxel data, a child reference set of
voxel data, a first condition reference set of voxel data for a
first physiological condition, and a second condition reference set
of voxel data for a second physiological condition.
23. The system as defined in claim 21 further comprising a
microprocessor for automatically selecting the viewing protocol
from the plurality of viewing protocols based on at least one of an
identify of a user, a profession of the user, an organization
affiliation of the user, and a content of the new set of voxel
data.
24. A computer program product for use on a computer, the computer
program product comprising: a recording medium; and, means recorded
on the recording medium for instructing the computer system to
perform the steps of: a) storing a reference set of views for a
reference set of voxel data according to the viewing protocol,
wherein each view in the reference set of views is defined by a set
of geometrical parameters defined with respect to a reference
coordinate system of the reference set of voxels; b) defining a
registration mapping from the reference set of voxel data to the
new set of voxel data; and, c) using the registration mapping to
map each view in the reference set of views to the new set of voxel
data to provide the set of views for the new set of voxel data
according to the viewing protocol.
25. The computer program product as defined in claim 24 wherein the
reference set of views comprises at least two views.
26. The computer program product as defined in claim 25 wherein
step a) comprises reviewing different views of the reference set of
voxel data to select and store the reference set of views.
27. The computer program product as defined in claim 24 wherein the
means recorded on the recording medium is further operable to
instruct the computer system to perform the step of generating a
set of views for a 2nd new set of voxel data according to the
viewing protocol after generating the set of views in step c), by
d) defining a registration mapping from the reference set of voxel
data to the 2nd new set of voxel data; e) using the registration
mapping to map each view in the reference set of views to the 2nd
new set of voxel data to provide the set of views for the 2nd new
set of voxel data according to the viewing protocol.
28. The computer program product as defined in claim 24 wherein the
viewing protocol further comprises a set of geometrical constraints
including at least one of i) a shape constraint for preserving a
view shape during step c), ii) an angular constraint for preserving
a viewing angle during step c), iii) a spacing constraint for
preserving internal relative spacing during step c), iv) an angular
view-to-view constraint for preserving an angular relationship
between different views in the set of views during step c), and v)
a spacing view-to-view constraint for preserving a spacing
relationship between different views in the set of views during
step c); wherein step c) further comprises transforming the set of
views to comply with the geometrical constraints to generate a
corrected set of views according to the viewing protocol.
29. The computer program product as defined in claim 24 wherein, in
step c), the registration mapping is a volume-to-volume
registration mapping.
30. The computer program product as defined in claim 24 wherein the
reference set of voxel data is obtained from at least one of (i) an
anatomical atlas for storing the reference set of voxel data
segmented into labeled regions; (ii) a person being scanned; (iii)
a composite obtained from a person being scanned multiple times;
and, (iv) a composite obtained from a plurality of people being
scanned.
31. The computer program product as defined in claim 24 wherein
each view in the reference set of views is defined by a set of
non-geometrical parameters defined with respect to the reference
set of voxels, wherein the set of non-geometrical parameters
comprises at least one attribute selected from the group consisting
of contrast, transparency and color.
32. The computer program product as defined in claim 31 wherein, in
step c), using the registration mapping to map each view in the
reference set of views to the new set of voxel data comprises, for
each view in the reference set of views, defining a corresponding
view in the new set of voxel data by, i) mapping a geometrical
parameter in the set of geometrical parameters to a corresponding
geometrical parameter of a new set of geometrical parameters
defined with respect to a new reference coordinate system of the
new set of voxel data, and ii) modifying each non-geometrical
parameter defined with respect to the set of voxel data based on
the new set of voxel data.
33. The computer program product as defined in claim 24 wherein the
means recorded on the recording medium is further operable to
instruct the computer system to perform the step of, before step
b), selecting the viewing protocol from a plurality of viewing
protocols, wherein step a) comprises storing a plurality of
reference sets of views for a plurality of reference sets of voxel
data, wherein each reference set of views in the plurality of
reference sets of views is of a corresponding reference set of
voxel data in the plurality of reference sets of voxel data, and
defines a corresponding viewing protocol in the plurality of
viewing protocols.
34. The computer program product as defined in claim 33 wherein the
plurality of reference sets of voxel data includes at least one of
a male reference set of voxel data, a female reference set of voxel
data, an adult reference set of voxel data, a child reference set
of voxel data, a first condition reference set of voxel data for a
first physiological condition, and a second condition reference set
of voxel data for a second physiological condition.
35. The computer program product as defined in claim 25 wherein the
means recorded on the recording medium is further operable to
instruct the computer system to perform the step of automatically
selecting the viewing protocol from the plurality of viewing
protocols based on at least one of an identify of a user, a
profession of the user, an organization affiliation of the user,
and a content of the new set of voxel data.
Description
[0001] The present invention relates generally to
computer-generated images and, more particularly, to a method for
storing and comparing views of three-dimensional images and
sequences of three-dimensional images.
BACKGROUND OF THE INVENTION
[0002] Physicians are faced with an increased need to more
efficiently interpret large volumetric image datasets produced by
medical imaging scanners utilizing technologies such as computed
tomography (CT) or magnetic resonance imaging (MRI). These imaging
scanners typically produce a series of cross-sectional images or
slices in each patient session for viewing by the physician.
Sequences of two-dimensional slices may be combined to produce
information about a volume corresponding to an area of the
patient's anatomy. Whereas, in the past, viewing the slice images
of the image volume individually, side-by-side, was sufficient, the
increased number of slices routinely acquired with modern
imagers--often approaching 1,000 slices/exam--renders such methods
impractical. As well, the increased number of possible diagnostic
purposes for which such volumetric data is useful further compounds
the problem. Physicians are therefore increasingly using 3D
rendering methods, such as pixel averaging, maximum intensity
projection (MIP), and color/opacity compositing for examining the
volume contents.
[0003] In current 3D viewing systems, physicians are typically
presented multiple views, often rendered in viewport windows and
tabulated in a "collage" layout, with their geometry
cross-referenced to assist the user in understanding their
relationship to each other and to the patient's anatomy. The
selection of appropriate parameter values for optimal 3D views is
often very time-consuming.
SUMMARY
[0004] In accordance with an aspect of an embodiment of the
invention, there is provided a method of automatically generating a
set of views for a new set of voxel data according to a viewing
protocol. The method comprises a) storing a reference set of views
for a reference set of voxel data according to the viewing
protocol, wherein each view in the reference set of views is
defined by a set of geometrical parameters defined with respect to
a reference coordinate system of the reference set of voxels; b)
defining a registration mapping from the reference set of voxel
data to the new set of voxel data; and, c) for each view in the
reference set of views, using the registration mapping to map each
view in the reference set of views to the new set of voxel data to
provide the set of views for the new set of voxel data according to
the viewing protocol.
[0005] In accordance with another embodiment of the invention,
there is provided a system for automatically generating a set of
views for a new set of voxel data according to a viewing protocol.
The system comprises a memory for storing the new set of voxel data
and the viewing protocol; and means for performing the steps of a)
storing a reference set of views for a reference set of voxel data
according to the viewing protocol, wherein each view in the
reference set of views is defined by a set of geometrical
parameters defined with respect to a reference coordinate system of
the reference set of voxels; b) defining a registration mapping
from the reference set of voxel data to the new set of voxel data;
and, c) for each view in the reference set of views, using the
registration mapping to map each view in the reference set of views
to the new set of voxel data to provide the set of views for the
new set of voxel data according to the viewing protocol.
[0006] In accordance with a yet further embodiment of the
invention, there is provided a computer program product for use on
a computer. The computer program product comprises a recording
medium; and, means recorded on the recording medium for instructing
the computer system to perform the steps of: a) storing a reference
set of views for a reference set of voxel data according to the
viewing protocol, wherein each view in the reference set of views
is defined by a set of geometrical parameters defined with respect
to a reference coordinate system of the reference set of voxels; b)
defining a registration mapping from the reference set of voxel
data to the new set of voxel data; and, c) for each view in the
reference set of views, using the registration mapping to map each
view in the reference set of views to the new set of voxel data to
provide the set of views for the new set of voxel data according to
the viewing protocol.
BRIEF DESCRIPTION OF THE DRAWINGS
[0007] The skilled person in the art will understand that the
drawings, described below, are for illustration purposes only. The
drawings are not intended to limit the scope of the applicant's
teachings in any way.
[0008] FIG. 1 provides examples of slab geometries and their
respective sampling grids.
[0009] FIG. 2 is a flowchart illustrating a method of creating and
saving a 3D viewing protocol in accordance with an aspect of an
embodiment of the invention.
[0010] FIG. 3 is a flowchart illustrating a method of applying a
viewing protocol using volume-to-volume registration in accordance
with a further aspect of an embodiment of the invention.
[0011] FIG. 4 is a block diagram illustrating a computer system in
accordance with an embodiment of the invention.
[0012] FIG. 5 is a flowchart illustrating a method of applying an
atlas-based viewing protocol using volume-to-volume registration in
accordance with a further aspect of an embodiment of the
invention.
DETAILED DESCRIPTION OF THE INVENTION
[0013] Examples of 3D rendering geometries are illustrated in FIG.
1. These include: i) a planar slab 10, in which an extruded plane
of a given thickness (typically 0-10 mm) is generated; ii) a curved
slab, in which an extruded curve of a given thickness is
generated); and iii) simulated camera views, in which a virtual
camera is placed at a given orientation and location (inside or
outside the data volume) and an image is computed using orthonormal
or perspective projection geometry. In the case of curved slab
rendering, the extrusion may be of a fixed curve along a linear
path 20 (2D), a fixed curve along a curved path 30 (2.5D), or a
variable curve along a curved path 40, 50 (3D). With simulated
camera views, the view may optionally contain cutting planes and
masks to help uncover anatomical regions that would otherwise be
hidden from view of the virtual camera.
Creating a Saved Viewing Protocol
[0014] Referring now to FIG. 2, one embodiment of the process of
creating a saved viewing protocol is illustrated in greater detail.
A patient is scanned using a medical imaging scanner 201, such as a
CT or MRI system. The medical imaging scanner produces a sequence
of two-dimensional image slices. The sequence of 2D image slices is
combined to synthesize image data in a third-dimension, to produce
three-dimensional volume data. Alternatively, the medical imaging
scanner may provide complete 3D volume data P.
[0015] The volume data is represented as a set of volume cells, or
voxels, representing data values at grid locations in
three-dimensional space. Voxels in the data volume are
geometrically associated with a reference coordinate system (RCS)
P, chosen for convenience. For example, each voxel intensity value
may have associated Cartesian coordinates (x.sub.P, y.sub.P,
z.sub.P). Alternatively, a suitable substitute coordinate system
may be used, such as, e.g., a cylindrical coordinate system.
[0016] To view the contents of the volume, the user is presented
with a collage of one or more cross-referenced views 211. A
rendering algorithm 203, controlled by a set of rendering
parameters 204 prescribing the rendering operation, generates each
view. The parameters may include the rendering type (e.g., average,
MIP, compositing), the geometry type (e.g., flat slab, curved slab,
or simulated camera), the geometrical coordinates and angles
associated with the given geometry type, and additional optional
parameters. Optional parameters may include, for example, color,
contrast, transparency, volumetric masks or thresholding ranges to
highlight specific features in a volume.
[0017] Some of the parameters may be set to change automatically
over a given range to create animation effects. For example, the
rendering algorithm may animate the output image to produce a
rotating effect or to progressively unmask or cut away anatomical
elements.
[0018] The rendering techniques describe here, and their
controlling parameters, are well known in the art and are embodied
in commercial visualization workstations marketed by CT and MRI
scanner manufacturers, such as Philips (www.medical.philips.com),
GE (www.gehealthcare.com), Siemens (www.medical.siemens.com), as
well by dedicated workstation developers such as Vital Images
(www.vitalimages.com) and TeraRecon (www.terarecon.com).
[0019] Using an input device 205, such as a mouse or keyboard, the
user may modify some or all of the rendering parameters until the
views in the collage best show the anatomy of interest. The user
then saves in step 206 the collage description 207 to a persistent
storage medium, such as a computer hard drive. The collage
description 207 contains the layout of the views 211 within the
collage viewing area 210, and, for each view, the parameters 204
that specify the rendering operation for generating that view.
[0020] The user may further change the layout of the views 211 in
collage viewing area 210, adjust its contents by changing rendering
parameters 204, and again saving in step 206 the new collage
description 207 to the persistent storage medium to create a saved
protocol 208, consisting of multiple collage descriptions 207. This
process may be repeated as many times as necessary.
[0021] The result is a sequence of collages forming a streamlined
review protocol 208 preferred by the user when reviewing the type
of anatomy shown in the data volume P. The data volume P from which
the collages were generated is saved in step 220, forming a
protocol consisting of a data volume P and a sequence of collage
descriptions 207, each containing views described by rendering
parameters 204 relating to data volume P. The use of data volume P
is described in greater detail below, with regard to FIG. 3. By
retaining and using the collage descriptions 207 and data volume P,
the system is able to quickly and reliably generate a sequence of
pre-determined renderings in a familiar order and in relation to
familiar objects to facilitate diagnosis by a clinician.
[0022] Moreover, saved protocols may be generated for various
classes of patients to facilitate easier comparison. For example, a
clinician may save a protocol for male patients, female patients,
children or for specific medical conditions.
[0023] For example, when reviewing a CT angiography exam of the
head and neck, the user may create a sequence of collages, each
presenting a region of possible abnormalities: left carotids, right
carotids, middle cerebral artery, Circle of Willis, etc. In each
collage, the views of the region are optimized for a quick
detection of blood flow abnormalities. For example, the left
carotid collage may present a simulated camera view of the neck
from the left side, a curved slab of 5 mm thickness through the
middle of the carotid, and a small 1 mm flat slab view programmed
to scroll automatically along the length of the carotid,
perpendicular to its centerline, at 1 mm increments.
Applying the Viewing Protocol using Volume Registration
[0024] The viewing protocol created for volume P may subsequently
be used with other data volumes, such as a current data volume C,
acquired from another patient or from the same patient at a
different time. As described above, the particular profile used can
be selected from among many different protocols. The selection of
protocol P from amongst the set of available protocols can be done
either manually by selection from a descriptive list, or
automatically, by the computer executing programmed decision rules
that select the most suitable protocol to apply based on the
identity and/or profession of the user (eg, radiologist or surgeon)
and/or organizational affiliation of the user (eg, department A or
department B) and/or the contents of the data and its associated
descriptors (eg, anatomical region, clinical indications for the
study) and/or the nature of the patient (male or female, adult or
child, physiological condition).
[0025] Referring now to FIG. 3, there is shown a flowchart
illustrating the steps involved in a method according to an aspect
of an embodiment of the invention. The method involves applying a
saved viewing protocol 208 and rendering parameters 204 using
volume-to-volume registration obtained in step 301. The method
involves automatic data-volume to data-volume registration, in
which two or more data volumes are correlated. Many volume to
volume registration algorithms are currently known in the art, as
surveyed in the book "Handbook or Biomedical Image Analysis Volume
III: Registration Methods", edited by Suri, Wilson, Laxminarayan,
2005, the contents of which are included here by reference. The
multi-resolution algorithms described next is similar to the one
described in the paper "A Generic Framework for Non-Rigid
Registration Based on Non-Uniform Multi-Level Free-Form
Deformations", Julia A. Scnablel et al, in Proc. of MICCAI
2001.
[0026] The registration of the two volumes generates a mapping T(x)
of any coordinate in RCS P to a corresponding anatomical location
in RCS C, or a P.fwdarw.C coordinate mapping 302. A possible
implementation of the volume-volume registration uses a non-rigid
mapping, which is a combination of global and local
transformations:
T(x)=T.sub.global(x)+T.sub.local(x)
[0027] The global transform is computed first, using an affine
transformation algorithm, to compensate for pose differences
between data volumes by maximization of normalized mutual
information. Next, the local transform is computed to facilitate
compensation for local deformations, using a spline-based approach.
Deformation is modeled as free-form deformation using B-splines,
which can be written as the 3D tensor product of 1D cubic
B-splines:
T local = t = 0 3 m = 0 3 n = 0 3 B l ( u ) B m ( v ) B n ( w )
.phi. i + l , j + m , k + n ##EQU00001##
[0028] Where .PHI. denotes the control points which parameterize
the transform.
[0029] The optimal transform is found by minimizing a cost function
associated with both the global and local transform parameters. The
cost function can comprise two components representing competing
goals: maximizing voxel value correlation in the regions around
each of the control points and maintaining smoothness and
regularity in the transform. Control points may be defined in
volume P in a suitable arrangement, for example, on a regular grid
of a desired resolution (e.g., 1 cm spacing).
[0030] Using the P.fwdarw.C coordinate mapping 302, the geometrical
parameters of each view in P are mapped 305 to coordinate space C.
Where a parameter describes an angle or a directional unit vector,
it is converted into two sets of coordinates (i.e., the base and
tip of a vector), which are then mapped from P to C to provide the
equivalent vector in C.
[0031] Viewing assumptions may be explicitly or implicitly encoded
in the rendering parameters 204. In one example, flat slab viewing
geometry is defined by a rectangular section of a plane
representing one boundary of the slab, and a thickness over which
sampled values would be processed to generate the output image. The
rectangular section may be further defined by the P coordinates of
its corners. In a second example, a curve may be specified by a
sequence of equidistant positions along the curve, each position
having a specific coordinate in volume P. The requirement that the
positions be equidistant may be specified in the curve descriptor
or implicitly assumed by the program reading the descriptor from
file.
[0032] As organs have different sizes and shapes among different
individuals, coordinate mapping will rarely preserve distances and
angles. Thus, the mapped locations in volume C may violate
geometrical constraints associated with the view type (e.g.,
equidistant curve points). Thus, a processing step 306 may be
applied following the mapping to compute parameters that observe
the geometrical constraints, yet remain as close as possible to the
transformed views.
[0033] In one example, locations on the planar rectangular section
in volume P do not form a planar rectangular section in volume C. A
set of new corner coordinates may be produced by sampling the
section in volume C in a rectangular grid, transforming each grid
location to volume P, computing the closest plane in volume P using
bilinear regression, then selecting a rectangular section of the
plane whose corners are the closest to the corners of the
transformed grid. The thickness of the slab, in mm, remains
unchanged.
[0034] In a second scenario, the mapped curve locations in volume C
are not equidistant. Therefore, the mapped curve in volume C may be
re-interpolated using a spline algorithm and re-sampled at new
equidistant positions.
[0035] Geometrical constraints may also apply to the spatial
relationship between multiple views in the same collage, such as
relative angles and distances. Such constraints are best handled in
a combined manner that best reflects the intended use of the views.
For example, in a commonly used arrangement, the collage contains 3
views of mutually-perpendicular flat slabs intersecting at a
location of interest, for example at the center of a lesion. In
such a case, it is preferable to represent in the collage the
geometry of all 3 views together as consisting of an intersection
point and two perpendicular orientation unit vectors specifying the
orientation of one of the planes. Upon retrieval, the intersection
point and the ends of the two orientation unit vector (with their
bases placed at the intersection point) are mapped to the C
coordinate system in step 305. After the mapping, the vectors may
not be of unit length and may not be perpendicular to each other.
The constraints can be reapplied to the two vectors using well
known linear algebra techniques, for example by using 3 cross
products to generate a new set of mutually-perpendicular vectors
similar to the original ones, which are then normalized to a unit
length.
[0036] Non-geometrical rendering parameters, such as color,
contrast and transparency may be passed on to the rendering step
unmodified, or, optionally, they may be recomputed based on the
contents of data volume C. For example, a thresholding value for
hiding voxels in a 3D projection may be required to allow
visualization of bright objects. This value may be specified either
as a fixed value, in which case it is simply passed on unmodified,
or it may be specified as some function of the data value histogram
(e.g., at the 80.sup.th percentile), in which case it is
recomputed.
[0037] A collage of multiple views in the protocol is presented to
the user by computing new rendering parameters 307 for each of the
views, as described above, and rendering them 308 in the collage
viewing area 210 according to the collage's layout description. The
user can step through the collages stored in the protocol to obtain
a complete presentation of the anatomy of the given body
region.
Atlas-Based Protocols
[0038] In some advanced applications, the data volume of interest
can be registered to an atlas volume, in which data regions have
been labeled or tagged. The atlas can serve, for example, as a
model for segmenting volume P into anatomically meaningful regions.
In such a case, the atlas volume itself may serve as a reference
volume, eliminating the need for saving dataset C itself as an
additional reference for the viewing protocol.
[0039] An alternative atlas-based method for saving a protocol is
described in FIG. 5. For clarity, the same reference numerals, with
200 added, are used to designate steps of the atlas-based method of
FIG. 5 relative to analogous steps of the method of FIG. 3. For
brevity, the description of FIG. 3 is not repeated in relation to
FIG. 5. Data volume P is registered to atlas volume A. The result
of the registration, mapping P.fwdarw.A, is then used to map the
relevant rendering parameters, established in step 509, to the
atlas coordinate space. Constraints are then applied to preserve
the viewing parameters' integrity, as described above, and the
views are finally saved, in step 508, with reference to the atlas
volume A, which is already in the database.
[0040] The process of later applying the viewing protocol to a new
data volume C is identical to the one shown in FIG. 3, except that
the atlas data volume A takes the place of protocol data volume
P.
[0041] In another embodiment, similar in implementation to the
atlas-based one, the reference data volume A, whether labeled or
not, may be advantageously generated as a composite of individual
datasets, obtained from multiple people. The compositing process
averages the anatomical variations in a population to create a more
typical arrangement of anatomy, enabling the registration of new
datasets to the reference dataset to be more robust than in the
case the reference dataset came from a single individual with some
unusual anatomical features. Another advantage of compositing
multiple datasets with partial spatial overlap is increased
anatomical coverage, enabling the support region for registration
to be potentially larger than in the case of a single dataset. This
advantage applies not only when the composited datasets were
obtained from different persons, but also when they were obtained
from the same person at different times.
[0042] A variety of methods for compositing individual datasets
into a single population-averaged dataset are known in the art. For
example, see "A Population-Average, Landmark- and Surface-based
(PALS) atlas of human cerebral cortex", Van Essen, Neuroimage. 2005
November 15;28(3):635-62, and "Groupwise Construction of Appearance
Models using Piece-wise Affine Deformations", Cootes et al, Proc.
of the British Machine Vision Conference, September 2005.
[0043] Referring to FIG. 4, there is illustrated in a block diagram
a computer system 400 for automatically generating sets of views in
accordance with an embodiment of the invention. As shown, the
computer system 400 can comprise a display 410, memory 412,
processor 414 and input/output device 416. Optionally, all of these
components may be provided on a single workstation. Alternatively,
some of the elements may be located offsite. For example, in an
alternative embodiment the display 410 and input/output device 416
can be provided in a workstation computer used by a user, such as a
medical professional, while some components at least of the memory
412 and processor 414 are provided offsite. In such an embodiment,
these components of the memory 412 and processor 414 could, for
example, be remotely accessed by the user via the workstation and
the internet.
[0044] In operation, memory 412 is operable to store a viewing
protocol module (VPM) 418. This viewing protocol module (VPM) 418
can comprise a single reference set of views or a plurality of
reference sets of views. Each reference set of views i) is for a
corresponding reference set of voxel data; ii) can comprise at
least two views, and iii) defines a particular protocol. Thus,
where multiple reference sets of views are included in the viewing
protocol module (VPM) 418, a user or the system can choose between
multiple viewing protocols. A user, for example, could choose
between viewing protocols using the input/output device 416 and a
protocol selection module (PSM) 420. Alternatively, protocol
selection module (PSM) 420 can be operable to automatically select
a particular viewing protocol from among the plurality of viewing
protocols based on factors, such as, for example, the identity of
the user, the profession of the user, an organizational affiliation
of the user, and a content of a new set of voxel data being
considered.
[0045] The plurality of reference sets of voxel data, on which
corresponding reference sets of views are defined to provide the
plurality of viewing protocols, may comprise distinct reference
sets of voxel data for each of the following groups:
[0046] 1) an adult male reference set of voxel data;
[0047] 2) an adult female reference set of voxel data;
[0048] 3) a boy reference set of voxel data;
[0049] 4) a girl reference set of voxel data;
[0050] 5) a first condition reference set of voxel data for a first
examination type (such as Cardiac Angiography); and,
[0051] 6) a second condition reference set of voxel data for a
second physiological condition (such as Colongraphy).
[0052] According to an aspect of an embodiment of the present
invention, the reference set of views for a reference set of voxel
data according to a particular viewing protocol are selected by a
user, using the input/output device 416, when viewing different
views on the display 410. The particular views selected are then
stored in the reference set of views for that protocol. The views
may be obtained from an anatomical atlas (M) 421, also stored in
memory 412, or from a person being scanned. Each view in each
reference set of views can be defined by a set of geometrical
parameters defined with respect to a reference coordinate system of
the reference set of voxels, and by a set of non-geometrical
parameters defined with respect to the reference set of voxels. The
set of non-geometrical parameters can comprise attributes such as
contrast, transparency and color.
[0053] Once the viewing protocol module (VPM) 418 has been
developed, the system 400 is ready to pre-process a set of views
for a new set of voxel data, based on a particular viewing protocol
selected by the user via the protocol selection module (PSM) 420,
or by the protocol selection module (PSM) 420 itself. Of course,
where the viewing protocol module (VPM) 418 comprises but a single
reference set of views for a reference set of voxel data to define
only a single viewing protocol, no selection of a particular
viewing protocol is required.
[0054] When the new set of voxel data is provided, then, as
described above, a registration mapping module (RMM) 422 determines
a registration mapping from the reference set of voxel data to the
new set of voxel data. In one embodiment, this registration mapping
is a volume-to-volume registration mapping. In establishing the
registration mapping, the registration mapping module (RMM) 422 can
map each view in the reference set of views to the new set of voxel
data by, for each view in the reference set of views, 1) mapping a
geometrical parameter in the set of geometrical parameters to a
corresponding geometrical parameter to provide a new set of
geometrical parameters defined with respect to a new reference
coordinate system of the new set of voxel data; and 2) modifying
each non-geometrical parameter defined with respect to the
reference set of voxel data based on the new set of voxel data.
[0055] As described above, applying the registration mapping in
this manner may distort the new set of views. To help to filter out
some of these distortions, a correction module (CM) 426 is operable
to generate a corrected set of views according to the viewing
protocol, by applying a set of geometrical constraints (GC) 424
stored in memory 410. Again, as described above, the set of
geometrical constraints can comprise: 1) a shape constraint for
preserving a view shape in the new set of views for the new set of
voxel data; 2) an angular constraint for preserving a viewing angle
in the new set of views for the new set of voxel data; 3) a spacing
constraint for preserving internal relative spacing in the new set
of views for the new set of voxel data; 4) an angular view-to-view
constraint for preserving an angular relationship between different
views in the new set of views for the new set of voxel data; and 5)
a spacing view-to-view constraint for preserving spacing
relationships between different views in the new set of views for
the new set of voxel data. By this means, the view correction
module (CM) 426 is operable to generate a corrected set of views
according to the viewing protocol selected.
[0056] Other variations and modifications of the invention are
possible. For example, it will be appreciated by those of skill in
the art that aspects of embodiments of the present invention can
also be embodied in instructions stored on a suitable recording
medium, which instructions could subsequently be used to configure
a computer system to implement an embodiment of the present
invention. All such modifications or variations are believed to be
within the sphere and scope of the invention as defined by the
claims appended hereto.
* * * * *